SYMPOSIUM SERIES No.

150 # 2004 IChemE

FLASH POINTS OF AQUEOUS SOLUTIONS OF

FLAMMABLE SOLVENTS

G.R. Astbury1, J. Bugand-Bugandet2, E. Grollet2 and K.M. Stell

Avecia Limited, P.O. Box 42, Blackley, Manchester M9 8ZS
1
Current address: Health & Safety Laboratory, Harpur Hill, Buxton,
Derbyshire, SK17 9JN
2
Ecole des Mines d’Albi, Campus Jarlard, 81013 Albi, Cedex 09, France

Many fine chemicals and pharmaceutical intermediates are processed using

flammable solvents which are soluble in water. The disposal of such solutions of sol-
vents as waste can be problematical, particularly as dilute aqueous solutions can be
expensive to incinerate. Disposal to an effluent treatment plant can be hazardous if
the solution is above its flash point. Hence the availability of the flash point of
aqueous solutions of flammable solvents would be useful for selecting an appropriate
disposal route. Whilst such data of actual measurements of aqueous solutions are
valuable, a predictive method would also be useful, and the data of three solvents
have been compared to simple theoretical models, to allow prediction to be under-
taken using readily available spreadsheets or calculating tools, rather than using dedi-
cated specialist software for physical property prediction. The decision to avoid
proprietary prediction software was taken deliberately to make the method easily
applicable to users in small and medium enterprises (SMEs) who may not have
access to such specialised programmes. The correlations obtained were good, with
actual flash points being within 5 K of the predicted values. Tables of measured
values are presented for 18 common solvents.

INTRODUCTION
The measurement of flash point of solutions can be difficult and time consuming. The most
common flash-point apparatus, the Pensky-Martens closed-cup flash point tester1, is suit-
able for pure substances, and is not suitable for a two component mixture, as this will
change in composition as the testing proceeds, thus giving a false high flash-point. There-
fore an equilibrium method is required, such as the Setaflash Closed Cup method2, or the
standard single-component method must be modified to an equilibrium method. Once flash
point data is available for several aqueous mixtures of any one solvent, it is comparatively
simple to draw a curve through the known data points and interpolate to any desired value.
However, obtaining the initial flash-point data is time consuming and requires access to the
flash-point testing apparatus. Therefore a simple method of predicting the flash-point of a
solution of solvent in water was required, using vapour-liquid equilibrium data and pure
component flash-point data readily available in the literature. The approach of using a
simple prediction method using the minimum of data is ideal for SMEs who often do
not have access to sophisticated physical property data prediction software, and only
have access to the open published literature.

1
SYMPOSIUM SERIES No. 150 # 2004 IChemE

FLASH POINT
The flash point of a liquid is the temperature at which the liquid has sufficient vapour
pressure to just form a flammable atmosphere in equilibrium with the liquid. Clearly
for a single component, such as acetone, the vapour concentration at the flash point
corresponds directly with the lower flammable limit. This will also apply to a multi-
component mixture where all the components are flammable.
However, in a binary mixture where one of the components is flammable and the
other is water, there is inevitably water vapour present as well as the flammable vapour.
Where there is a small quantity of water vapour present, the flammable material dominates
and the lower limit concentration will be reached at just a little over the temperature of the
flash point of the pure flammable liquid.
Where the water is present at a high concentration, there are two factors involved.
Firstly, the concentration of the flammable vapour is depressed by the effect of the water in
the solution, and the water vapour itself reduces the partial pressure of the oxygen in the air
within the flash-point apparatus. Hence there will be a concentration where despite there
being enough flammable vapour present to be above the lower flammable limit, there is
insufficient partial pressure of oxygen present. Thus it can be difficult to determine
whether the mixture is above its flash point but has insufficient oxygen to burn, or
whether it is simply below its flash point.

METHOD OF MEASURING FLASH POINT

The flash points were determined using a Sommer and Runge PMA2 Automatic Pensky-
Martens Closed Cup Flash Point Tester supplied by Gallenkamp. The procedure used was
similar to that described in the relevant British3, American4 and European5 Standards.
However, as the standard Pensky-Martens Closed Cup Flash Point Tester is designed
for single components only, a modified procedure was used.
The standard method for single-component mixtures heats the sample at a rate of
1 K per minute whilst being stirred. A small flame is dipped into the flash-point tester
cup periodically, and if there is sufficient vapour, a flash of flame occurs. The testing by
dipping the flame into the cup is initiated typically between 20 K and 40 K below the
expected flash point. The temperature at which this occurs is recorded as the flash
point. As a single component liquid will give off a vapour of exactly the same composition
as the liquid, any vapour lost between successive tests will have no effect on the compo-
sition of the remaining liquid.
Where a two or more component liquid is tested in the same fashion, the vapour will
be richer in composition with the more volatile component. In the case of flammable sol-
vents, this will usually be the solvent rather than the water. Hence if the standard test is
carried out, starting some 20 K below the expected flash point, the loss of the more volatile
component will alter the composition significantly. In the case of the solvent being
the more volatile component, the recorded flash point will be far higher than the true
flash point.

2
SYMPOSIUM SERIES No. 150 # 2004 IChemE

In order to avoid this potential error, the technique was modified to estimate the
flash point as normal so as to initially determine an approximate value for the flash
point. This approximate value can then be used to set the flash point tester so that it
will start testing just below the approximate flash point using a fresh sample. Successive
approximations lead to the situation where a fresh sample will flash immediately on testing
at one temperature, yet will not flash if tested only 2 K below. This allows the Pensky-
Martens method to approximate to an equilibrium method for multi-component mixtures.

SELECTION OF SOLVENTS
The solvents selected, 18 in all, were chosen on the basis that many were commonly used
in the fine chemicals and pharmaceutical sector, although some are frequently used in
other sectors as well. Since solvents which are insoluble in water exhibit a substantially
constant flash-point irrespective of the concentration of the two-phase solvent-water
mixture, only completely miscible and partially miscible solvents were selected. All
were flammable, and a range of pure component flash points was selected to cover the
range from 2408C to þ578C with a reasonable coverage of the range. Details of the sol-
vents are contained in Table 2. In the table, the solubility is described as “total” where it is
completely miscible, “partial” where the solubility is not complete, but is more than
10% v/v, and “slight” where the solubility is less than 10% v/v. Note that some of the
solvents classed as “slight” form two-phase liquid mixtures over a large part of the
range, be it at one end or in the middle.

EXPERIMENTAL
Solutions of each solvent were prepared by mixing water and the solvent together, with the
quantities being measured by volume. This was both for convenience, and to minimise the
time that the solvent was evaporating, as this would affect the final composition. The com-
position in volume-volume concentrations can be converted to mass-mass concentration
by using the equation:

100
(Vs
rs )
W¼ (1)
(Vs
rs ) þ (Vw
rw )

Where:
W ¼ weight concentration of solvent %(w/w).
VS ¼ volume of solvent
rS ¼ density of solvent
Vw ¼ volume of water
rw ¼ density of water
For all practical purposes, the density of water can be taken as unity, so the equation can be
reduced to W ¼ (Vs
rs )=(Vs
rs ) þ Vw . The samples were prepared by mixing at

3
SYMPOSIUM SERIES No. 150 # 2004 IChemE

ambient temperature, and were then tested for flash point, using the modified testing pro-
cedure outline above. The results are contained in Appendix 1 for reference.
There were some inconsistencies which gave difficulties during the testing. The first of
these were where the liquids formed two phases. Over the two-phase region, there was
little variation in flash point with composition, although some liquids, such as methyl
ethyl ketone (MEK, 2-Butanone), exhibit retrograde solubility, in that a single phase sol-
ution at ambient temperature forms two separate phases when heated6. Where this has
occurred, it is noted in the Appendix. Similarly, there are cases where two phases at
ambient temperature form a single phase when heated to the flash point, but this has no
affect on the flash-point since it is single phase at the test temperature.
Further complications occurred for tert-butanol (2-methyl-2-propanol) which is a
solid at the flash point of 118C, and melts at approximately 268C. To determine the
flash point of the pure material, molten liquid was put into the Setaflash apparatus, and
cooled to well below the solidifying temperature before testing. The addition of 10% by
volume of water to the pure molten liquid lowered the freezing point to below ambient,
so this allowed testing to be undertaken more readily for the water-containing samples.
Finally, the tetrahydrofuran solutions were difficult to test as, upon cooling, those
containing less than 10% water formed ice crystals at sub-zero temperatures, altering
the liquid phase composition. This continued until the concentration of the water exceeded
about 90%, by which time the flash-point had reached þ68C, and sub-cooling of the liquid
below 08C was not necessary to determine an accurate flash point.

PREDICTION OF FLASH POINT

For a single component, the vapour concentration at the flash-point corresponds to the
lower flammable limit. Hence when the flash point is known, it is easy to use vapour press-
ure data from the literature6,7 to determine the vapour concentration by partial pressure.
Although there are many commercial software packages available to predict
physical properties, many such programmes are not available to small and medium enter-
prises (SMEs), and many are temporal, in that the current version becomes obsolete in a
few years and will no longer run on today’s computer platforms. Therefore a method was
sought which did not rely on specific commercial software, so as to be future-proofed.
However, it was considered that a spreadsheet would be useable to solve the equations
rapidly and plot the flash-point versus concentration curves, and as such would not be
reliant on any specific software package.
Whilst there is little available data on measured flash-points of aqueous solutions, it
was thought desirable to have a method where the flash point of any composition of an
aqueous solution could be predicted from readily available data, typically the pure
liquid flash point and some data on the vapour liquid equilibrium available from the litera-
ture8,9. This data would be used to predict when the vapour composition of the flammable
component reaches the lower flammable limit, as this would correspond to the flash-point.
As much data is isobaric, some correction would have to be applied to take account of the
variation in temperature between the data available and the predicted flash point10.

4
SYMPOSIUM SERIES No. 150 # 2004 IChemE

Where vapour-liquid equilibrium data are not available, it would be necessary to

have some predictive method to determine the vapour liquid equilibrium data. This
again would require a prediction method, and the method of van Laar11 was deemed to
be simple, and require the minimum of data, although the use of other more sophisticated
methods such as the Wilson, NTRL (Non Random Two Liquids) or UNIQUAC
(UNIversal QUasi-chemical Activity Coefficients), would be more accurate but generally
require more data than is commonly available. Clearly where more data are available, it
would be better to use a more sophisticated and hence more accurate method, but the
objective of this work was to find a simple predictive method that required the absolute
minimum of data, which would generally be available in common data sources such as
Lange’s Handbook of Physics and Chemistry12, Perry’s Chemical Engineers’ Hand-
book13, or International Critical Tables6. Tables of azeotropic systems are available in
papers by Horsley14,15.

SYSTEMS WHERE x-y-t DATA ARE AVAILABLE

The calculation method is best illustrated by an example calculation, as this allows easy
use simply by substituting data values into the method. The data available for methanol
is suitable. Selected smoothed vapour-liquid equilibrium data16 for methanol-water at
730 to 750 mm Hg is as in Table 1 below.
The vapour pressure at any temperature can be calculated from the simple Antoine
formula:

0:05223
log10 pmm ¼  A+B (2)
T

Table 1. Vapour – liquid equilibrium

for methanol –water at 760 mm Hg

Mole % Methanol in

Liquid Vapour Temp8C

0 0 100
2 13.4 96.4
4 23.0 93.5
6 30.4 91.2
8 36.5 89.3
10 41.8 87.7
15 51.7 84.4
20 57.9 81.7

5
SYMPOSIUM SERIES No. 150 # 2004 IChemE

Table 2. Properties of solvents selected

Solvent Flash Point 8C Boiling Point 8C Solubility

Acetaldehyde 240 21 Total

Tetrahydrofuran (THF) 221 65.5 Total
Methyl Acetate 218 57.5 Slight
Methyl Ethyl Ketone (MEK) 26 80 Partial
Acetonitrile 5 81.5 Total
2,3-Butanedione 8 88 Partial
Methanol 9 65 Total
2-Propanol 14 82.4 Total
Tert.Butanol 12 83 Total
p-dioxane 12 101 Total
Methyl IsoButyl Ketone (MIBK) 15 118 Slight
Allyl Alcohol 22 97 Total
1-Propanol 23 97 Total
2-Butanol 22 98 Slight
Isobutanol 30 108 Slight
Cyclohexylamine 28 134 Total
1-Butanol 37 117 Slight
Morpholine 35 129 Total
Acetic Acid 39 116 Total
Ethylene Glycol Isopropyl Ether 46 143 Total
N,N-Dimethyl Formamide (DMF) 58 153 Total

where pmm is the vapour pressure in mm Hg, the constants A and B for methanol are 38324
and 8.8017 respectively17 and T is the absolute temperature. Since the flashpoint of metha-
nol at 760 mm Hg is 98C, this corresponds to a vapour pressure of

0:05223
log10 pmm ¼ 
38324 þ 8:8017 ¼ 1:7086: [ pmm ¼ 50:5 mm Hg
273:2 þ 9

and this corresponds to a lower flammable limit of (50.5/760)
100 ¼ 6.65% v/v which
compares well with the literature value18 of 6.7% v/v.
If a 4 mole % solution of methanol is considered, then the vapour above this solution
will contain 23 mole % at the boiling point of 93.58C, i.e. a partial pressure of
(23/100)
760 ¼ 174.8 mm Hg. As the lower flammable limit is 6.7 mole %, it corre-
sponds to the partial pressure of 50.5 mm Hg. Therefore, it is necessary to estimate the
temperature at which the partial pressure of the methanol above the 4 mole % solution
is 50.5 mm Hg. This is a simplification which relies on the relative volatility between
the methanol and the water not changing appreciably with temperature. This is a

6
SYMPOSIUM SERIES No. 150 # 2004 IChemE

reasonable assumption when the vapour pressures are low, and far removed from the criti-
cal temperature or pressure, which is the case for methanol and water. The temperature at
which the partial pressure of the methanol above the solution reaches 50.5 mm Hg is deter-
mined from the pure component vapour pressure of 174.8 mm Hg at the boiling point of the
solution for which the vapour-liquid equilibrium data is obtained. At 93.58C, the vapour
pressure of methanol is 2203 mm Hg, so the temperature at which the solution has a
partial pressure of 50.5 mm Hg and just becomes flammable would have a pure-component
vapour pressure of (50.5/174.8)
2203 ¼ 636 mm Hg. This corresponds to a temperature
of 60.58C obtained by solving the Antoine equation. As the solution contained 4 mole %, it
is necessary to convert to mass concentration, using the equation:

100
(Ms xs )
W¼ (3)
Ms xs þ Mw xw

so the mass concentration for 4 mole % of methanol (mol. weight 32) and water (mol. weight
18) corresponds to W ¼ (100
(32
0.04))/(32
0.04 þ 18
0.96) ¼ 6.90% by
weight. This same calculation can be repeated for all the known vapour-liquid equilibrium
data available, to give a range of calculated flash points versus concentration. These are
plotted in Figure 1 along with the measured values amended to a weight-weight basis.
As methanol is completely miscible with water, the same method was used with a
highly non-ideal system of 1-butanol and water. This exhibits partial solubility and forms

Figure 1. Methanol flash point variation with concentration

7
SYMPOSIUM SERIES No. 150 # 2004 IChemE

Figure 2. 1-Butanol flash point variation with concentration

an azeotrope. However, using the published vapour-liquid equilibrium data8, a graph of

flash-point versus concentration can be plotted using the same method. This is shown in
Figure 2, and again shows a good correlation.

SYSTEMS WHERE x-y-t DATA ARE NOT AVAILABLE

In this case, as vapour-liquid equilibrium data are not available, it is necessary to use a
predictive method to estimate the vapour-liquid equilibrium, and then apply the same
method as above. As another example, this has been carried out for a solution of
p-dioxane. We chose dioxane as there was no vapour-liquid equilibrium data readily avail-
able, but both azeotropic data and Antoine equation vapour pressure data were available.
The mixture is reported14 to form an azeotrope containing 18% w/w water, boiling at
87.88C. As it is necessary to determine the vapour concentration in equilibrium with the
liquid, a predictive method such as van Laar’s method can be used where the liquid is
single phase. The method of van-Laar, whilst not particularly accurate, has the advantage
of requiring only one experimental point since it is a two-constant equation. In this
example, the data point of the azeotrope is used to estimate the vapour-liquid equilibrium.
The azeotrope is very useful as it is easier to obtain an accurate vapour composition than
with a non-azeotrope.

8
SYMPOSIUM SERIES No. 150 # 2004 IChemE

Firstly, the vapour-liquid equilibrium data is estimated. The activity coefficients at

the azeotropic concentration are calculated first as:

P
gs ¼ (4)
p0s

and

P
gw ¼ (5)
p0w

Since ys and xs are equal at the azeotrope, then the two values can be substituted into the
van Laar equations to give:
 
xw ln gw 2
a ¼ ln gs 1 þ (6)
xs ln gs

and
 2
xs ln gs
b ¼ ln gw 1þ (7)
xw ln gw

from this the final equations for activity coefficient are:

a
ln gs ¼   2 (8)
xs
1 þ ba
xw

and
b
ln gw ¼   2 (9)
b xw
1þa
xs

The vapour composition at an estimated boiling temperature can now be calculated from
the expressions:

gs p0s xs
ys ¼
P
gw p0w xw
yw ¼ (10)
P

9
SYMPOSIUM SERIES No. 150 # 2004 IChemE

Figure 3. p-Dioxane flash point variation with concentration

If the sum of the mole fractions of the solvent and water does not equal 1.0, then the
boiling temperature is adjusted until the sum of the mole fractions does equal 1.0. This
now gives the vapour-liquid-equilibrium/temperature data required for the estimation
of flash point, and the calculation is continued as before where the vapour liquid equili-
brium data were available. The calculations were carried out on a spreadsheet, and the
calculated flash points for dioxane-water are displayed in Figure 3.

DISCUSSION
The data presented in Appendix 1 for the 18 solvents is actual measured data. The deter-
mination of the flash point itself did present some difficulties in a practical sense. Firstly, as
explained above, the need to use an equilibrium method necessitated modifying the stan-
dard test methods given in the American, British or European standards. The modification
to the method does mean that the Standards are not adhered to strictly, and hence anyone
not familiar with the problems of two-component mixtures could easily be misled into
measuring a false high flash point. At the limit, a flash may not be detected at all, since
the loss of the volatile component can be significant unless a “go/no-go” test criterion
is applied with a fresh sample used for each test. Whilst this is a valid extension to the
test method, it is not in accordance with the Standards, and hence must be detailed with
the reported results.

10
SYMPOSIUM SERIES No. 150 # 2004 IChemE

The lower strength solutions presented practical problems, in that when the solution
boiled in the standard cup, the boiling liquid overflowed from the cup, and entered the
heating element and the apparatus itself. On the early tests, this damaged the equipment,
so a modified cup was used with a annular guard lip around the cup to catch the liquid
boiling over. Again, this does not strictly conform to the standards, but as it is external
to the geometry of the cup itself, it does not affect the result. Standard single components
flashed at the same temperature in both the guarded and standard cups. A photograph of the
guarded and standard cups is shown in Figure 4.
Further complications occurred with two samples, as indicated above. Firstly, tert-
butanol and dioxane are unusual in that they exhibit a flash point below the melting point.
Clearly, this cannot be tested in a standard Pensky-Martens flash point apparatus, as there
is a stirrer which cannot be used for a solid. Hence the Setaflash equilibrium tester was
used. The solid material was melted, and the pure liquid placed in the cup and cooled
until it was frozen. The cup was then allowed to warm up, and the flash-point was deter-
mined for the pure component. For the aqueous solutions, the molten liquid was added to
water, and allowed to dissolve. The addition of a small quantity of water was sufficient to
lower the freezing point to below the flash point, and so the Pensky-Martens tester could
then be used for the subsequent tests.
The testing of tetrahydrofuran was complicated by the formation of ice-crystals.
Here, the solution exhibited a rapid change in solubility of the water with respect to
temperature, and ice crystals formed in the solution upon cooling below zero. Since
the aim of this work was to determine the flash point of aqueous solutions, principally

Figure 4. Flash point apparatus modified cup (left) and standard cup (right)

11
SYMPOSIUM SERIES No. 150 # 2004 IChemE

for determining whether effluent streams were flammable, the work only considered
those solutions where the flash point was above 08C, i.e. where the tetrahydrofuran
concentration was less than 10% v/v. This behaviour is unusual in that most solutions
of solvents with water act as an antifreeze, as in ethanol, methanol and ethylene
glycol, all of which depress the onset of formation of ice crystals well below zero
Celsius.
When undertaking the prediction of flash point, it was decided that a simple
approach would be preferable, as if this gave acceptable results it would be easy and
quick and would only require reference to standard readily available text-books and refer-
ence works, which would generally be available to SMEs. The use of the two-coefficient
Antoine equation for vapour pressure prediction was used as the data set in International
Critical Tables6 was based on this form. Obviously Antoine equations with three or more
constants can be used with this method, but the overall accuracy is obviously dependant on
the accuracy of the vapour liquid equilibrium data as well.
The use of the van Laar prediction of vapour-liquid equilibrium is only the simple
option chosen here – other methods would probably be equally suitable, but no work was
undertaken to establish whether more accurate predictive methods for vapour-liquid equi-
librium would give more accurate predictions of flash point. This work could be under-
taken in future if time permits. The work of Liaw et al uses a more complex model, but
does so only where both substances are flammable.

CONCLUSIONS
. Closed-cup flash point data are presented for aqueous mixtures of 18 common
solvents.
. A predictive method using vapour-liquid equilibrium data and pure-component vapour
pressure data has been developed which gives acceptable results for predicting the
flash point of aqueous solutions.
. Where only vapour pressure data and a single azeotrope composition are available, a
further predictive method has been shown to provide an acceptable estimation of the
flash-point of an aqueous solution.
. The simple approach used will be of benefit to small and medium sized enterprises who
may not have access to sophisticated predictive tools and software.
. The methods above are suitable for calculating using standard spreadsheets on
computers.
. The method does not rely on specific commercially available software, as hand calcu-
lation methods can also be used.

NOMENCLATURE
a arbitrary constant in the van Laar equation
b arbitrary constant in the van Laar equation
M molecular weight

12
SYMPOSIUM SERIES No. 150 # 2004 IChemE

P atmospheric pressure
p vapour pressure, mm Hg
V volume
W weight %
x mole fraction in liquid
y mole fraction in vapour

Greek
g activity coefficient
r density

Superscripts
0
pure component

Subscripts
s solvent
w water

APPENDIX 1
FLASH POINT DATA FOR AQUEOUS SOLVENT MIXTURES
Notes:
All data obtained from in-house testing at Blackley
NFTB ¼ No Flash to Boil

¼ Tested on a Setaflash flash-point tester as the material is a solid at the indicated
temperature

Acetaldehyde
Acetic Acid
Concentration %v/v Flash Point +28C
100 240 Concentration Flash Point
5 þ14 %v/v +28C
2 31 100 39
1 45 90 50
0.5 66 88 NFTB

13
SYMPOSIUM SERIES No. 150 # 2004 IChemE

Acetonitrile 1-Butanol

Concentration %v/v Flash Point +28C Concentration %v/v Flash Point +28C
100 5 100 37
5 45 90 42
3 60 # #
2 NFTB 8 44
7 46
6 47
5 50
Allyl Alcohol 40 53
2 64
Concentration %v/v Flash Point +28C 1 NFTB
# Two-phase between
100 22 8 and 90% v/v
90 27
80 29
70 30
60 31 2-Butanol
50 32
40 33 Concentration % v/v Flash Point +28C
30 35
20 39 100 22
10 49 90 27
5 61 # #
4 NFTB 10 34
5 43
2 64
1 NFTB
# Two-phase between
2,3-Butanedione (Diacetyl) 8 and 90% v/v

Concentration %v/v Flash Point +28C

100 8
35 11 iso-Butanol (2-Methyl-1-Propanol)
30 18
20 19 Concentration %v/v Flash Point +28C
10 29
100 30
5 41
90 34
2 56
1 NTFB
(Continued)

14
SYMPOSIUM SERIES No. 150 # 2004 IChemE

iso-Butanol (2-Methyl-1-Propanol) Cyclohexylamine

Concentration %v/v Flash Point +28C Concentration %v/v Flash Point +28C
# # 60 55
8 34 50 55
5 40 40 55
3 48 30 56
2 55 20 56
1 NFTB 10 55
# Two-phase between 5 55
8 and 90% v/v 3 58
2 NFTB

t-Butanol
N,N-Dimethyl Formamide
Concentration %v/v Flash Point +28C
Concentration %v/v Flash Point +28C
100 12
90 15 100 58
80 17 90 70
70 18 88 73
60 18 86 76
50 19 84 79
40 20 83 NFTB
30 21
20 24
10 32
5 43
1,4-Dioxane
2 57
1 NFTB
Concentration %v/v Flash Point +28C

100 12
Cyclohexylamine 90 14
80 15
Concentration %v/v Flash Point +28C 70 15
60 17
100 28 50 20
90 36 40 24
80 44 30 30
70 52 20 38

(Continued) (Continued)

15
SYMPOSIUM SERIES No. 150 # 2004 IChemE

1,4-Dioxane Methyl Acetate

Concentration %v/v Flash Point +28C Concentration %v/v Flash Point +28C
10 50 6 11
5 64 5 13
3 76 4 16
2 NFTB 3 22
2 30
1 46
2-Isopropoxyethanol (Ethylene glycol 0.5 63
isopropyl ether) 0.4 NFTB
#
Setaflash
Concentration %v/v Flash Point +28C
100 46
90 56 1-Propanol
80 61
70 64 Concentration %v/v Flash Point +28C
65 NFTB
100 23
90 28
80 30
Methanol 70 31
50 31
Concentration %v/v Flash Point +28C 30 32
20 35
100 9 10 43
75 19 5 52
50 27 4 54
25 41 3 57
10 61 2 68
5 NFTB 1 NFTB

Methyl Acetate 2-Propanol

Concentration %v/v Flash Point +28C Concentration %v/v Flash Point +28C
#
100 218 100 14
8 7 90 18

(Continued) (Continued)

16
SYMPOSIUM SERIES No. 150 # 2004 IChemE

2-Propanol Morpholine

Concentration %v/v Flash Point +28C Concentration %v/v Flash Point +28C
80 19 80 49
70 21 70 57
60 21 60 69
50 21 55 75
40 23 50 NFTB
30 25
20 30
10 41
5 50
2 65 MIBK (4-Methyl-2-Pentanone)
1 NFTB
Concentration %v/v Flash Point +28C
100 15
# #
2-Butanone (Methyl Ethyl Ketone) 2 20
1 31
Concentration %v/v Flash Point +28C 0.5 43
0.3 57
100 26 0.2 NFTB
# # # Insoluble above
30 22 2%
20 þ2
10 11
5 20
2 36
1 50 Tetrahydrofuran
0.5 64
0.3 NFTB Concentration %v/v Flash Point +28C
100 221
# #
10 þ6
Morpholine 4 22
3 26
Concentration %v/v Flash Point +28C 2 33
1 46
100 35
0.5 62
90 41
#Sample formed ice
when cooled
(Continued)

17
SYMPOSIUM SERIES No. 150 # 2004 IChemE

REFERENCES
1. BS EN 22719:1994, “Methods of Test for Petroleum and its Products. Petroleum Products
and Lubricants. Determination of Flash Point. Pensky-Martens Closed Cup Method ”.
2. ASTM D3278 – 96, “Standard Test Methods for Flash Point of Liquids by Setaflash Closed-
Cup Apparatus”.
3. BS EN 22719: 1994, “Methods Of Test For Petroleum And Its Products. Petroleum Pro-
ducts And Lubricants. Determination of Flash Point. Pensky-Martens Closed Cup
Method ”.
4. ASTM D93– 99c, “Standard Test Method for Flash Point by Pensky-Martens Closed
Tester”.
5. Test A9: “Flash Point”, Directive 92/69/EEC, Official Journal of the European Commu-
nities, L383, 29th December 1992.
6. International Critical Tables, 1929, 3: 387, McGraw Hill Book Co.
7. Reid, R.C., Prausnitz, J.M., Sherwood, K.T., 1987, Properties of gases and liquids,
McGraw-Hill Book Co.
8. Chu, J.C., Wang, S.L., Levy, S.L., Paul, R., 1956, Vapor-Liquid Equilibrium Data, J.W.
Edwards Publisher Inc.
9. Chu, J.C., Getty, R.J., Brennecke, L.F., Paul, R., 1950, Distillation Equilibrium Data,
Reinhold Publishing Corp., New York
10. Liaw, H-J., Lee, Y-H., Tang, C-H., Hsu, H-H, Liu, J-H., 2002, A mathematical model for
predicting the flash point of binary solutions, J. Loss Prev. Proc. Ind., 15, 429– 438.
11. van Laar, J.J., 1913, Z. Phyzik Chem., 72; 723, and 83; 599.
12. Lide, David R. (Ed), 2000, CRC Handbook of Chemistry and Physics, 81st Ed.
13. Perry, R.H., & Green, 1984, Chemical Engineers’ Handbook, McGraw Hill.
14. Horsley, L.H., 1947, Table of Azeotropes and Non-azeotropes, Ind. Eng. Chem. Anal.
Chem., 19; 508– 600
15. Horsley, L.H., 1949, Table of Azeotropes and Non-azeotropes, Ind. Eng. Chem. Anal.
Chem., 21; 831– 873
16. Perry, J.H., 1963, Chemical Engineers’ Handbook, 4th Edn., page 13 – 5
17. International Critical Tables, 1929, 3: 215, McGraw Hill Book Co.
18. Coward, H.F., and Jones, G.W., 1952, Limits of Flammability of Gases and Vapors, U.S.
Bureau of Mines Bulletin 503.

18